Re: [PATCH] srcutiny: Add lockdep and early-boot checks to synchronize_srcu_atomic()

[Bradley Morgan <brads@mainlining.org>]

On 12 September 2026 00:12:25 BST, "Paul E. McKenney" <paulmck@kernel.org>
wrote:
>On Fri, Sep 11, 2026 at 10:18:04PM +0100, Bradley Morgan wrote:
>> On 11 September 2026 17:09:13 BST, Kunwu Chan <kunwu.chan@gmail.com>
>wrote:
>> >Add the RCU_LOCKDEP_WARN and early-boot check that the Tree version
>> >gained, srcu_lock_sync() was already present.
>> >
>> >Suggested-by: Zqiang <qiang.zhang@linux.dev>
>> 
>> Ok.
>> 
>> Reviewed-by: Bradley Morgan <brads@mainlining.org>
>
>Applied, thank you!
>
>> >Signed-off-by: Kunwu Chan <kunwu.chan@gmail.com>
>> >---
>> > kernel/rcu/srcutiny.c | 6 ++++++
>> > 1 file changed, 6 insertions(+)
>> >
>> >diff --git a/kernel/rcu/srcutiny.c b/kernel/rcu/srcutiny.c
>> >index 99f8bfd98b04..76411562366f 100644
>> >--- a/kernel/rcu/srcutiny.c
>> >+++ b/kernel/rcu/srcutiny.c
>> >@@ -361,6 +361,12 @@ void synchronize_srcu_atomic(struct srcu_struct
>*ssp)
>> > 
>> > 	srcu_lock_sync(&ssp->dep_map);
>> > 
>> >+	RCU_LOCKDEP_WARN(lockdep_is_held(ssp),
>> >+			 "Illegal synchronize_srcu_atomic() in same-type SRCU read-side critical section");
>> >+
>> >+	if (rcu_scheduler_active == RCU_SCHEDULER_INACTIVE)
>> >+		return;
>> >+
>> 
>> Wow!
>
>If you think that this one is fun, take a look at the implementation of
>synchronize_rcu() in kernel/rcu/tiny.c.  ;-)


void synchronize_rcu(void)
{
    RCU_LOCKDEP_WARN(...);
    preempt_disable();
    WRITE_ONCE(rcu_ctrlblk.gp_seq, rcu_ctrlblk.gp_seq + 2);
    preempt_enable();
}

Hah!

>
>							Thanx, Paul
>
>> > 	if (IS_ENABLED(CONFIG_PREEMPTION))
>> > 		synchronize_rcu(); // Needed for RCU Tasks Trace to imply RCU grace period.
>> > 				   // And in Tiny RCU, it is near zero cost and doesn't block.
>> >
>> 
>> --- Thanks!
>>
>https://lore.kernel.org/all/EE579805-42F2-4C58-B752-F28779EEB717@grrlz.net/

--- Thanks!
https://lore.kernel.org/all/EE579805-42F2-4C58-B752-F28779EEB717@grrlz.net/